MODELING THE CORTICAL VISUAL PATHWAYS USING ARTIFICIAL NEURAL NETWORKS
thesisposted on 03.12.2021, 18:08 authored by Zhixian HanZhixian Han
Although in conventional models of visual information processing, object identity and spatial information are processed separately and independently in ventral and dorsal cortical visual pathways respectively, some recent studies have shown that information about both object’s identity (of shape) and space are present in both visual pathways. However, it is still unclear whether the presence of identity and spatial information in both pathways have functional roles or not. In a recent study (Han & Sereno, in press), we have tried to answer this question through computational modeling. Our simulation results suggested that two separate cortical visual pathways for identity and space (1) actively retain information about both identity and space; (2) retain information about identity and space differently; (3) that this differently retained information about identity and space in the two pathways may be necessary to accurately and optimally recognize and localize objects. However, in these simulations, there was only one object in each image. In reality, there may be more than one object in an image. In this master’s thesis, I have tried to run visual recognition simulations with two objects in each image. My two object simulations suggest that (1) the two separate cortical visual pathways for identity and space (orientation) still retain information about both identity and space (orientation) when there are two objects in each image; (2) the retained information about identity and space (orientation) in the two pathways may be necessary to accurately and optimally recognize objects’ identity and orientation. These results agree with our one object simulation results.